Type 2 Diabetes Polygenic Score in Addition to Clinical Factors for Prediction of Diabetes Incidence in an Indigenous American Population

Background: Type 2 diabetes (T2D)-associated variants, derived from genome-wide association studies (GWAS), have largely been reproducible across populations, but there is limited information on how polygenic scores (PS) based on these variants add to clinical factors for predicting diabetes incidence, particularly in non-European populations. Methods: A longitudinal study of diabetes was conducted in an Indigenous study population native to the Southwestern U.S. At each exam, a 75-g oral glucose tolerance test was administered with measurement of fasting and 2-hour plasma glucose (FPG, 2hPG). Genotypes were available from a GWAS for 2783 participants aged ≥ 20 years without diabetes at baseline, with imputed genotypes based on whole genome sequence data from 296 individuals from the population. PSs for participants were constructed and weighted using genome-wide significant variants in a T2D GWAS meta-analysis in European-descent populations (n=245; Mahajan 2018). Survival analyses were conducted to predict diabetes incidence in 2 models: (1) age, sex, BMI, FPG; (2) age, sex, BMI, FPG, T2D PS. Area under the receiver-operating characteristic curve (AUC) was calculated using a nonparametric method for survival analysis. Results: 903 cases (32.5%) of diabetes occurred during 25,110 person-years of follow-up. The hazard ratio for T2D PS, controlled for age, sex, BMI and FPG, was 1.27 per SD (P=1.5×10-12). Cumulative incidence at 10 years was 0.076 in the lowest quartile for the prediction score based on clinical factors alone and 0.509 in the highest quartile. For the model including PS, corresponding values were 0.075 and 0.524. AUC was 0.710 for the model with clinical factors alone and 0.721 for the model including T2D PS. Conclusion: Weighted T2D PSs based on 245 variants were informative in predicting T2D incidence in Indigenous Americans; however, the improvement in prediction beyond clinical factors alone was modest. Disclosure L.E. Wedekind: None. S. Kobes: None. W. Hsueh: None. L. Baier: None. W.C. Knowler: None. A. Mahajan: None. M.I. McCarthy: Advisory Panel; Self; Illumina. Consultant; Self; Eli Lilly and Company, Novo Nordisk Inc., Zoe Global Ltd. Employee; Self; Genentech, Inc. Research Support; Self; AbbVie Inc., Merck & Co., Inc., Sanofi-Aventis, Servier, Takeda UK. R.L. Hanson: None.